Steam: From a Mine Pump to a General Prime Mover

Before there was any use for a general source of power, there was a specific and expensive nuisance: coal mines fill with water.

Dig below the water table and groundwater seeps in continuously. Bail it out with horses turning a winding gear and the cost rises with depth, because the water must be lifted further and there is more of it. By the early eighteenth century British mines were reaching depths where the horse teams cost more than the coal was worth, and shafts were being abandoned while still productive.

Key idea: The first steam engines were not built to power anything. They were built to lift water out of holes, and they were installed at coal mines because that is where both the problem and the free fuel were. Everything the technology became followed from improvements to a machine designed for that single job.

Note the spacing. More than half a century separates Newcomen's working engine from Watt's condenser, and another decade passes before rotary motion makes the engine useful for anything but pumping. Each step removed the constraint that the previous design had run into, and none of them was a leap from nothing.

Thomas Newcomen's engine of 1712 does not push with steam pressure. It uses steam to make a vacuum, and lets the atmosphere do the work.

1. Steam at barely above atmospheric pressure fills a large open-topped cylinder, letting the piston rise.
2. Cold water is sprayed into the cylinder. The steam condenses and the pressure inside collapses.
3. Atmospheric pressure pushes the piston down, which is the power stroke.
4. The piston is chained to one end of a rocking beam; the other end lifts the pump rod in the shaft.

The design is elegantly undemanding: it needs no high-pressure vessel, and the boilers and castings of 1712 could not have contained one safely. It also has a fatal thermodynamic flaw, which the next step makes explicit.

Definition: An atmospheric engine is powered by air pressure acting against a vacuum created by condensing steam, not by steam pressure. Every engine before Watt's later work, and Watt's own early ones, operated this way.

The flaw is in step 2. Spraying cold water into the cylinder condenses the steam, and it also chills the cylinder walls. On the next stroke, incoming steam has to reheat that whole mass of iron before it can fill the space, and that heat is then thrown away again by the next spray.

Every cycle heats and cools a large iron cylinder. The result was a thermal efficiency of slightly over half a percent: more than 99 percent of the coal burned went into repeatedly warming metal.

Gotcha: That number sounds like a machine nobody would build. It was built widely, and the reason is the argument from the previous lesson. At a coal mine, fuel had no market value at the pit-head, so efficiency was nearly irrelevant and only the capital cost and the water lifted mattered. Away from coalfields, where fuel had to be bought and carted, the same engine made no economic sense at all.

Stated that way it sounds obvious, and the fifty-seven years between the two designs is a caution about how obvious things look afterwards. Watt reached it while repairing a model Newcomen engine at the University of Glasgow and noticing how much steam a small cylinder consumed relative to its output.

Key idea: The invention was not a new source of power. It was the removal of a specific waste in an existing one, and the entire subsequent history of the technology is more of the same: identify where the energy is going and stop it going there.

The condenser was patented in 1769 and the first commercial Boulton and Watt engines were installed in 1776. The delay is instructive, because none of it was about the idea.

In practice: The middle row is the general lesson. A machine is limited by what can be manufactured to tolerance, and the precision to build one engine part often arrives from an unrelated industry. The third row is worth noting too: pricing an innovation on the cost it removes rather than the value it adds is a commercial technique still in use.

An engine that rocks a beam can pump. To drive machinery it must turn a shaft, and converting reciprocating motion into rotation took its own decade of work, arriving around 1781.

Two further pieces were needed before the rotation was usable:

• Smooth output. A single piston delivers power in pulses. A heavy flywheel stores energy through the stroke and releases it between, turning a series of shoves into steady rotation.
• Constant speed. A textile machine driven by a varying speed produces varying yarn. The centrifugal governor, a pair of weights spun by the output shaft that rise with speed and close the steam valve as they rise, holds the speed roughly constant without supervision.

Key idea: The governor is the historically significant part. It is a feedback controller: the output of the machine is measured and fed back to adjust its own input. It is among the first widely deployed automatic control systems, and James Clerk Maxwell's 1868 analysis of why governors sometimes oscillate is a founding paper of control theory.

Watt worked at low pressure and actively opposed high-pressure designs as dangerous, which they were: boilers exploded and people were killed.

But pressure is what makes an engine small. A low-pressure engine needs an enormous cylinder to produce useful force, so it is a building. Raise the pressure and the same force comes from a fraction of the volume, and the engine can be carried on a cart, mounted on a boat, or put on rails. Richard Trevithick and others pursued this from the early 1800s, as boiler-making and wrought iron improved enough to make it survivable.

In practice: This is the step that turned steam from a way of running a mine or a mill into general-purpose transport, and it depended on metallurgy rather than on any new idea about thermodynamics.

Water power had been used for centuries and was cheaper than steam for a long time after Watt. Its limitation was not cost but geography: a mill had to sit on a suitable fall of water, and the sites were finite, seasonal, and often nowhere near labour, materials or customers.

A steam engine can be installed anywhere coal can be delivered. The consequences follow directly:

• Factories relocated to towns and ports, where workers and shipping already were, instead of to river valleys.
• Production stopped being seasonal, since a boiler does not run dry in summer or freeze in winter.
• Scale stopped being capped by the flow of a particular river, and a bigger engine could always be bought.
• Rail and steamship freight collapsed the cost of moving heavy goods over land and sea alike.

Key idea: The steam engine's importance is often stated as raw power output, and that is the least interesting part. Its significance is that it decoupled the location of production from the location of energy, and that is a change in the geography of the economy rather than in its physics.

The steam engine's development follows a shape that recurs whenever a technology matters, and it looks nothing like the popular version.

1. It started as a point solution. Not a vision of powered industry, but a pump for a specific expensive problem.
2. It was terrible, and shipped anyway. Half a percent efficiency was acceptable because the economics of one location made it acceptable.
3. Improvement was constraint-chasing. Fuel waste, then manufacturing precision, then motion type, then size, each addressed only once it became the binding limit.
4. Adjacent industries set the pace. Cylinder boring came from cannon; safe high pressure came from better iron. The engine could not advance faster than what could be made.
5. The general-purpose use arrived last. Decades after the first working machine, and the thing that made it general was portability rather than power.

In practice: Steam also ran ahead of the science. Working engines preceded thermodynamics by a century, and Carnot's analysis of why they could not be perfectly efficient came in 1824, more than a hundred years after Newcomen's engine was pumping. Practice taught theory here, not the other way round.